How can I reliably determine uploaded file name in a Perl CGI script?

Question

The filename passed from an upload form to a Perl CGI (using CGI.pm) script depends on the client machine and may contain client-dependent path separators. Is there a reliable way of parsing the passed parameter to determine the filename (usually the last sub-string following the last path separator).

Answer 1

I've only ever experienced the issue with path separators when the client is using MSIE (implying Windows paths). I've used a fairly simple regular expression to handle that. However, you could extend the regular expression (or even split) to handle the most common path separators - '/', '\' and the occasional ':'.

Alternatively, you should be able to work out the filesystem type from the useragent string (perhaps using HTTP::DetectUserAgent or HTTP::BrowserDetect). Given that you could call the fileparse_set_fstype function of File::BaseName before parsing the file.

Something like:

use File::Basename;
use HTTP::BrowserDetect;

# ... get the filename into $upload_file and agent into $user_agent_string.

my $browser = HTTP::BrowserDetect->new($user_agent_string);

my $ostype;

$ostype = 'MSWin32' if $browser->windows;
$ostype = 'Unix' if $browser->unix;
# There are more tests available.

fileparse_set_fstype($ostype);

my $filename = basename( $upload_file);